Trump Administration Advises Banks to Test Anthropic's Mythos Model

Officials from President Donald Trump's circle, according to TechCrunch, are informally recommending that American banks consider testing Anthropic's new language model, Mythos. This message caused immediate bewilderment in the industry — shortly before this, the U.S.

Department of Defense officially added Anthropic to its list of companies presenting a risk to the supply chain. Two contradictory signals emanating almost simultaneously from different structures within the same administration have left the market in a quandary. Anthropic is one of the key players in the generative AI market, creator of the Claude family of models.

The company was founded in 2021 by former OpenAI employees led by Dario and Daniela Amodei. Anthropic is actively developing its enterprise direction, consistently positioning itself as a developer of safe and predictable artificial intelligence. Apparently, the Mythos model was created with the financial sector in mind — tasks with high requirements for reliability, solution traceability, and compliance with regulatory standards.

The supply-chain risk status assigned to Anthropic by the Department of Defense carries quite serious connotations. Historically, such designations were applied primarily to Chinese technology companies — Huawei, ZTE, and others — when American authorities had concerns about data security, foreign influence, or vulnerabilities in the hardware supply chain. Extending this practice to an American private AI developer became an extraordinary event and raised many questions.

The Pentagon provided no specific public grounds for including Anthropic in the list, which in itself adds to the concern. This is why the parallel informal promotion of Mythos among banks by officials close to the White House is perceived as an extremely contradictory signal. Formally, these two decisions come from fundamentally different departments with different priorities.

Financial regulators view the market through the lens of U.S. competitiveness and technological leadership, while the defense ministry views it through the lens of national security and supply chain risk management.

Often these priorities come into direct conflict. For banks that must make long-term strategic decisions about AI implementation, such uncertainty creates a real problem: it is unclear which signals to trust and whether today's encouragement will transform into tomorrow's restrictions. The American banking sector, meanwhile, is actively moving toward widespread AI adoption.

JPMorgan Chase, Bank of America, Goldman Sachs, and other major financial institutions are already deploying language models for automating analytics, customer service, risk management, and compliance checks. In this context, the choice of AI infrastructure provider is not merely a technical decision, but a deeply political one. Data breaches, model failures, or undesirable behavior in a banking context could have catastrophic regulatory and reputational consequences.

What is happening reflects a deeper systemic contradiction in American technology policy. On one hand — the desire to ensure the country's leadership in AI, stimulate commercial deployment, and not cede the market to competitors. On the other — growing concern about security, strategic dependence of government and critical infrastructure on private companies.

Anthropic finds itself at the epicenter of this conflict. For the company itself, the situation carries a double signal. Informal support from Trump administration officials could significantly accelerate penetration into the financial sector and strengthen commercial position.

However, the supply-chain risk status at the Pentagon could complicate or entirely block access to government and defense contracts — traditionally an important source of revenue for major American technology companies. Whether banks will follow the informal recommendations of officials and take on the regulatory risk associated with Anthropic will be shown in the coming time. But the very fact of public divergence of signals within one administration eloquently testifies: the regulatory landscape around AI remains deeply unstable, and any major investments in a specific vendor today inevitably carry political risk that was recently considered acceptable to ignore.